A radiation source positioning system based on swarm unmanned aerial vehicle cooperation and ad hoc network communication

Abstract

This invention provides a radiation source localization system based on swarm drone collaboration and self-organizing network communication, comprising: a signal frequency acquisition module, which constructs a self-organizing communication network for the target area through a swarm of drones to acquire preset features of the target signal source; a signal edge determination module, which, when any drone detects the target signal features, delineates the outline of the signal coverage area to determine the target area of ​​the target signal source; a target node determination module, which calculates a virtual center point of the target area based on the drone positions, uses the virtual center point as the origin, determines the positioning drones located in three orthogonal directions, and determines the spatial geometric relationship between the positioning drones and the target signal source; and a signal coordinate determination module, which determines the three-dimensional coordinates of the target signal source based on the spatial geometric relationship. This invention improves the localization accuracy of radiation sources, while also increasing swarm search efficiency, reducing system energy consumption, and enhancing anti-interference capabilities in complex environments.

Description

Technical Field

[0001] This invention relates to the field of radiation source localization technology, and in particular to a radiation source localization system based on swarm drone collaboration and self-organizing network communication. Background Technology

[0002] Radio positioning technology is widely used in military reconnaissance, spectrum monitoring, and emergency search and rescue. Currently, radiation source positioning methods based on UAV platforms mainly include direction finding, time difference positioning, and time-frequency difference positioning.

[0003] Direction finding positioning measures the angle of arrival of a target signal using an airborne antenna array. The target position is determined by multiple measurements from a single station or by cross-measurement from multiple stations. This method has relatively simple equipment and fast response, but its positioning accuracy decreases sharply with increasing distance and is significantly affected by multipath effects. Time difference positioning (TDPoS) uses multiple receiving stations to measure the time difference of arrival of the same target signal and calculates the position through the intersection of hyperbolas. This method offers high positioning accuracy, but it requires extremely strict time synchronization between receiving stations and at least three receiving stations to achieve two-dimensional positioning.

[0004] Traditional Time-Frequency Differential (TDOA) positioning systems typically use 2-3 drones for signal acquisition and positioning. When some drones fail to meet the positioning geometry due to flight attitude or environmental obstruction, they need to be temporarily dispatched to suitable locations, resulting in long response times and limited coverage. Time-Frequency Differential (TFD) positioning adds frequency difference information to TDOA, allowing simultaneous calculation of the target's position and velocity. However, this system requires extremely high accuracy in frequency synchronization, local oscillator stability, and the measurement of the receiver's own motion state, making system implementation complex.

[0005] Existing technologies, such as radio direction finding, radio navigation, radio wave ranging or velocity measurement, and radio wave reflection or re-radiation for positioning or presence detection, mostly rely on a small number of fixed or semi-fixed receiving nodes, lacking the ability for dynamic, large-scale node coordination. When facing wide-area, time-varying, and complex electromagnetic environments, traditional methods suffer from the following shortcomings: a small number of nodes leads to poor geometric configuration and limited positioning accuracy; inter-node communication relies on ground communication or pre-planned links, resulting in response lag; and they cannot achieve integrated operations of rapid search and precise positioning of radiation source areas.

[0006] Therefore, a radiation source positioning system is needed that can fully leverage the scale advantage of UAV swarms, possess self-organizing communication capabilities, and achieve wide-area rapid search and high-precision positioning. Summary of the Invention

[0007] The purpose of this invention is to provide a radiation source positioning system based on swarm drone collaboration and self-organizing network communication. By optimizing orthogonal direction nodes and using an event-triggered diffusion mechanism, combined with BeiDou high-precision timing and self-organizing network communication, the system improves the positioning accuracy of radiation sources, enhances swarm search efficiency, reduces system energy consumption, and strengthens anti-interference capabilities in complex environments.

[0008] To achieve the above objectives, the present invention provides a radiation source localization system based on swarm drone cooperation and ad hoc network communication, the system comprising: The signal frequency acquisition module is used to build a self-organizing communication network for the target area through a swarm of drones and continuously monitor and acquire preset characteristics of the target signal source; The signal edge determination module is used to determine the target area of ​​the target signal source by outlining the contour of the signal coverage area when any drone detects the preset characteristics of the target signal source. The target node determination module is used to calculate the virtual center point of the target area based on the positions of all UAVs participating in the boundary exploration, and to determine the positioning UAVs located in three orthogonal directions with the virtual center point as the origin, and to determine the spatial geometric relationship between the positioning UAVs and the target signal source. The signal coordinate determination module is used to determine the three-dimensional coordinates of the target signal source based on spatial geometric relationships.

[0009] Furthermore, building a self-organizing communication network for the target environment through drone swarms specifically includes: S11. Each drone is equipped with a communication module, which loads preset network parameters and initializes the network protocol stack after power-on. S12. Each UAV periodically broadcasts a beacon frame containing its own identifier and status on a preset channel, while simultaneously listening to the beacon frames of other UAVs. S13. Upon receiving the beacon frame, the drone records the identifier and signal strength of the sending drone and generates a neighbor table; S14. Each UAV exchanges routing information based on its neighbor table and through a self-organizing routing protocol, dynamically establishing a routing table and forming a self-organizing communication network. S15. During network operation, each UAV continuously monitors the link status of the self-organizing communication network. When a link interruption or node change is detected, route repair and topology reconstruction are automatically triggered.

[0010] Furthermore, each drone is equipped with a BeiDou positioning and timing unit to collect the signal frequency of the target signal source, specifically including: S21. Each UAV receives the time signal from the Beidou satellite and calibrates and corrects the frequency and phase of the local crystal oscillator in real time to obtain the real-time position coordinates of each UAV. S22. The drone swarm conducts a distributed wide-area search in the target area according to the preset coverage path. Each drone continuously monitors the spectrum environment of the target area and detects the preset characteristics of the target signal source in real time.

[0011] Furthermore, real-time calibration and correction of the local crystal oscillator's frequency and phase specifically includes: S31. Receive BeiDou pulses and time information sent by BeiDou satellites, and use BeiDou pulses as absolute time references; S32. Divide the high-frequency clock signal output by the local crystal oscillator to generate a local pulse; S33. Compare the time difference between the rising edges of the BeiDou pulse and the local pulse using a phase detector, and output the phase difference signal. S34. After processing the phase difference signal through a loop filter, a control voltage is generated; S35. Apply the control voltage to the control terminal of the voltage-controlled crystal oscillator to adjust the frequency and phase of the local crystal oscillator in real time.

[0012] Furthermore, the target region of the target signal source is determined, specifically including: S41. When any one or more drones detect that the strength of a target signal exceeds a preset threshold, it is determined that a suspected target signal has been detected, and a trigger signal is broadcast to nearby drones through a self-organizing communication network. S42. Nearby drones that receive the trigger signal move toward the triggering drone, forming a local high-density cluster around the suspected target signal. S43. Each UAV in the cluster measures the signal-to-noise ratio of the target signal it receives and aggregates the signal-to-noise ratio information to the server. S44. The server determines the location where the signal-to-noise ratio is lower than a preset threshold as the signal coverage edge based on the spatial distribution of the signal-to-noise ratio. S45. The UAV located at the edge of the signal coverage area hovers or scans, dynamically measures the signal boundary, gradually delineates the coverage outline of the signal source, and identifies the area inside the outline as the target area where the target signal source may exist.

[0013] Furthermore, the virtual center point of the target region is calculated, specifically including: S51. Obtain the position coordinates of all drones participating in the boundary exploration and the received signal-to-noise ratio of the same target signal measured by each drone; S52. Use the received signal-to-noise ratio value of each drone as the weight of each drone in the calculation, and add up the signal-to-noise ratio weights of all drones to obtain the total weight. S53. Multiply the longitude, latitude, and altitude values ​​of each UAV by the corresponding signal-to-noise ratio weights to obtain the weighted longitude, weighted latitude, and weighted altitude of each UAV. Sum the weighted longitude, weighted latitude, and weighted altitude of all UAVs to obtain the total weighted longitude, total weighted latitude, and total weighted altitude. S54. Divide the total weighted longitude by the total weight to obtain the longitude value of the virtual center point; divide the total weighted latitude by the total weight to obtain the latitude value of the virtual center point; divide the total weighted altitude by the total weight to obtain the altitude value of the virtual center point. S55. By integrating the longitude, latitude, and altitude values, the three-dimensional coordinates of the virtual center point are obtained.

[0014] Furthermore, the determination of spatial geometric relationships specifically includes: S61. Establish a local three-dimensional coordinate system with the virtual center point of the target area as the origin; S62. Obtain the position coordinates and attitude information of each UAV participating in the exploration in the global coordinate system; S63. Using the direction cosine matrix, the position coordinates and attitude information are transformed into a local three-dimensional coordinate system to obtain the three-dimensional direction vector of each UAV relative to the virtual center point. S64. Calculate the axial proximity between the three-dimensional direction vector of each UAV and the three orthogonal coordinate axes in the local three-dimensional coordinate system; S65. Based on axial proximity, the UAV closest to the three orthogonal axes is selected as the positioning UAV, and the spatial geometric relationship between the positioning UAV and the target signal source is determined.

[0015] Furthermore, determining the three-dimensional coordinates of the target signal source specifically includes: S71. Simultaneously intercept signals emitted by the target signal source by positioning the drone, and record the arrival timestamp of each intercepted signal; S72. Calculate the initial position of the target signal source based on the arrival timestamp; S73. Iteratively correct the initial position to obtain the three-dimensional coordinates of the target signal source.

[0016] Furthermore, the initial position of the target signal source is calculated, specifically including: S81. Obtain the three-dimensional coordinates of at least three reference UAVs and determine the reference UAV; S82. Obtain the arrival time difference of the target signal source to each reference UAV relative to the reference UAV, and convert the time difference into a distance difference; S83. Introduce an auxiliary variable, which is the distance from the target signal source to the reference UAV; S84. Using distance difference and auxiliary variables, the nonlinear TDOA hyperbolic equation system is transformed into a pseudo-linear equation system with respect to the three-dimensional coordinates of the target signal source and auxiliary variables. S85. The pseudo-linear equation system is solved using the first weighted least squares method to obtain the initial solution of the target coordinates and auxiliary variables; S86. Using the geometric constraint relationship between the three-dimensional coordinates of the target signal source and the auxiliary variables, construct the second weighted least squares equation, correct the initial solution, and obtain the initial position of the target signal source.

[0017] Furthermore, the initial position is iteratively corrected, specifically including: S91. Based on the initial position, construct a system of linear equations about the position correction using Taylor series, and solve them to obtain the correction amount for the current position. S92. Add the correction amount to the initial position to obtain the candidate new position, and calculate the error index of the candidate new position; S93. If the error index of the candidate new position is less than the error index of the initial position, then accept the candidate new position as the estimated position for the next iteration; otherwise, recalculate the candidate position. S94. When the correction amount reaches the preset iteration condition, stop the iteration and output the three-dimensional coordinates of the target signal source.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a radiation source positioning system based on swarm UAV collaboration and ad hoc network communication. By determining the center point of the target signal source area and selecting UAVs in three orthogonal directions to participate in positioning, it significantly optimizes the TDOA observation geometry, bringing the hyperbola intersection angle close to the optimal state, greatly reducing the geometric accuracy factor, and achieving sub-meter or even higher positioning accuracy under the same time measurement error. Simultaneously, an event-triggered diffusion mechanism enables a shift from blind traversal of the entire area to local adaptive focusing, avoiding ineffective flights and significantly shortening the target detection and tracking response time. The system also employs precise boundary exploration and node optimization, allowing only key nodes to perform high-frequency acquisition and high-computing calculations, while other nodes remain in low-power standby, effectively reducing overall swarm energy consumption and extending operational endurance. Combined with BeiDou high-precision time synchronization and ad hoc network, the system possesses decentralized resilience, allowing rapid reconstruction of the positioning configuration after some nodes fail, and maintaining nanosecond-level time synchronization and reliable positioning even in GPS denial or strong electromagnetic interference environments. In addition, the bee swarm observes the target from multiple angles simultaneously and corrects the target signal source position through algorithm iteration, effectively suppressing multipath effects and non-line-of-sight propagation errors, and significantly improving the stability and reliability of positioning results in complex environments such as densely populated urban areas. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of a radiation source localization system based on swarm drone collaboration and self-organizing network communication, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the process of constructing a self-organizing communication network for a target environment using a drone swarm, as provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the signal frequency acquisition process of the target signal source provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the real-time calibration and correction process for the frequency and phase of a local crystal oscillator provided in an embodiment of the present invention. Figure 5 This is a schematic diagram of the process for determining the target region of a target signal source according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the process for calculating the virtual center point of the target region provided in an embodiment of the present invention; Figure 7 A schematic diagram illustrating the process of determining spatial geometric relationships provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the three-dimensional coordinate process for determining a target signal source provided in an embodiment of the present invention; Figure 9 A schematic diagram illustrating the process of calculating the initial position of a target signal source according to an embodiment of the present invention; Figure 10 This is a schematic diagram of the iterative correction process for the initial position provided in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0022] Reference Figure 1 This embodiment provides a radiation source localization system based on swarm drone cooperation and ad hoc network communication. The system includes: The signal frequency acquisition module is used to build a self-organizing communication network for the target area through a swarm of drones and continuously monitor and acquire preset characteristics of the target signal source.

[0023] The signal edge determination module is used to determine the target area of ​​the target signal source by outlining the signal coverage area when any drone detects the preset characteristics of the target signal source.

[0024] The target node determination module is used to calculate the virtual center point of the target area based on the positions of all UAVs participating in the boundary exploration, and to determine the positioning UAVs located in three orthogonal directions with the virtual center point as the origin, and to determine the spatial geometric relationship between the positioning UAVs and the target signal source.

[0025] The signal coordinate determination module is used to determine the three-dimensional coordinates of the target signal source based on spatial geometric relationships.

[0026] In this embodiment, the signal frequency acquisition module first constructs a self-organizing communication network for the target area using a swarm of drones and continuously monitors the spectrum environment, acquiring preset characteristics of the target signal source. This allows for rapid discovery of suspected targets in a wide-area search, avoiding blind traversal. In the signal edge determination module, when any drone detects preset characteristics of the target signal source, the swarm immediately uses an event-triggered mechanism to collaboratively outline the signal coverage area, accurately determining the target region of the signal source. This achieves a shift from full-area search to local adaptive focusing, significantly shortening the positioning response time.

[0027] Subsequently, the target node determination module calculates the virtual center point of the target area based on the positions of all UAVs participating in the boundary exploration. Using this center point as the origin, it determines the positioning UAVs located in three orthogonal directions, constructing the optimal TDOA observation geometry configuration. This significantly reduces the geometric accuracy factor and lays the geometric foundation for high-precision positioning. Finally, the signal coordinate determination module, based on this spatial geometric relationship, uses the time difference of arrival data collected by the selected positioning UAVs to calculate the three-dimensional coordinates of the target signal source using the Chan-Taylor algorithm. This achieves sub-meter or even higher precision radiation source positioning under the same measurement error. Furthermore, by selecting only a small number of key nodes to participate in high-computation calculations, it effectively reduces the overall system energy consumption and extends the swarm's operational endurance.

[0028] Specifically, each UAV in this embodiment is equipped with the following components: the signal acquisition device uses an RF sensor that supports multi-band signal reception, with a sampling rate of not less than 500MS / s, a dynamic range of not less than 110dB, and signal strength indication (RSSI) and time of arrival marking functions. The BeiDou positioning and timing module uses a BeiDou-3 dual-mode timing chip, supports B1I / B2a / B3I frequency bands, has a positioning accuracy better than 10cm, a timing accuracy better than 20ns, and is equipped with an oven-controlled crystal oscillator (OCXO) to maintain local clock stability.

[0029] The self-organizing network communication equipment is a software-defined radio (SDR) module supporting multi-hop mesh networks, with a transmission rate of no less than 50Mbps, a communication radius of no less than 1km, and support for MAVLink / UAVCAN communication protocols. The power management system uses a high-energy-density lithium battery pack, supporting at least 2 hours of continuous flight and featuring adaptive power consumption adjustment. The ground server is equipped with computing resources such as a GPU, capable of receiving analog signal data collected by various UAVs (with the collection location and time stamp marked). It uses this data to calculate the time difference of arrival (TDOA) of signals collected by different UAVs and, based on the time stamp, location, and TDOA information, determines the target location of the radiation source.

[0030] As a preferred embodiment, a self-organizing communication network is constructed for the target environment using a drone swarm, specifically including: S11. Each drone is equipped with a communication module, which loads preset network parameters and initializes the network protocol stack after power-on.

[0031] S12. Each UAV periodically broadcasts a beacon frame containing its own identifier and status on a preset channel, while simultaneously listening to the beacon frames of other UAVs.

[0032] S13. Upon receiving the beacon frame, the drone records the identifier and signal strength of the sending drone and generates a neighbor table.

[0033] S14. Each UAV exchanges routing information based on its neighbor table and through a self-organizing routing protocol, dynamically establishing a routing table and forming a self-organizing communication network.

[0034] S15. During network operation, each UAV continuously monitors the link status of the self-organizing communication network. When a link interruption or node change is detected, route repair and topology reconstruction are automatically triggered.

[0035] In this embodiment, during the positioning task preparation phase, the system establishes a baseline and begins the search through system initialization and coordinated cruise. After the UAV swarm takes off, it first establishes an OFDM-based wireless IP Mesh network using an ad hoc network communication module. The physical layer employs OFDM modulation technology to support non-line-of-sight (NLOS) communication in complex urban or mountainous environments, and combines MIMO multi-antenna technology to improve channel capacity and link reliability. The network layer adopts a decentralized, multi-hop self-organizing routing protocol, where each node is both a data transceiver and a relay forwarder, dynamically discovering and maintaining the optimal transmission path. The system-level design follows the all-IP concept, supporting the TCP / IP protocol stack for seamless integration with existing wired or wireless networks, while also incorporating GPS / BeiDou positioning functionality to support location-based routing strategies.

[0036] The establishment of a mesh network (i.e., a self-organizing communication network) is an automated and distributed process. After powering on, nodes load preset network parameters and initialize the protocol stack. Then, they periodically broadcast "Hello" beacon frames on preset channels while simultaneously listening to the beacons of other nodes, building a neighbor table in parallel by recording the other party's ID and signal strength (RSSI). As neighbor relationships are established, nodes broadcast route requests (RREQ) and receive route responses (RREP) through on-demand routing protocols (such as AODV), gradually forming a complete routing table and constituting a dynamic mesh topology. After the network is established, topology maintenance is continuously performed. If a node moves, fails, or its signal is interrupted, neighboring nodes will quickly detect and trigger route repair (such as local repair or the source node re-initiating route discovery), automatically reconstructing to ensure continuous communication links.

[0037] Through the aforementioned technical means, the system achieves high reliability without a central hub; the failure of any node will not lead to overall paralysis, demonstrating extremely strong resilience. The network can be automatically established within seconds and achieve millisecond-level self-healing when nodes move or links are interrupted, perfectly adapting to the high-speed dynamic topology of the swarm. With the help of OFDM and multi-hop relays, the signal can bypass obstacles such as buildings to achieve beyond-line-of-sight communication, with a single node's air-to-ground communication distance reaching 30-50 kilometers, supporting coverage expansion of more than 9 hops. The system can provide continuous data transmission rates of 60Mbps to 100Mbps, supporting real-time transmission of multiple channels of high-definition video and sensor data, with low end-to-end latency, meeting the needs of real-time command and control. It also supports various topologies such as point-to-point, point-to-multipoint, chain, and mesh, and can be flexibly deployed according to mission requirements.

[0038] As a preferred embodiment, each UAV is equipped with a BeiDou positioning and timing unit to collect the signal frequency of the target signal source, specifically including: S21. Each UAV receives the time signal from the BeiDou satellite and calibrates and corrects the frequency and phase of the local crystal oscillator in real time to obtain the real-time position coordinates of each UAV.

[0039] S22. The drone swarm conducts a distributed wide-area search in the target area according to the preset coverage path. Each drone continuously monitors the spectrum environment of the target area and detects the preset characteristics of the target signal source in real time.

[0040] In this embodiment, the system utilizes an airborne BeiDou positioning and timing module to achieve space-based time synchronization. Specifically, it receives high-precision time signals from BeiDou satellites (such as 1PPS pulses per second and TOD time information) as an absolute time reference source. Using an internal phase-locked loop or discipline algorithm, it calibrates and corrects the frequency and phase of the airborne local high-stability crystal oscillator (such as an OCXO) in real time, thereby eliminating clock deviations and drift between UAV nodes. Simultaneously, each UAV acquires its initial position coordinates (latitude / longitude) and the absolute time of each signal file acquisition, providing a unified time reference and positional basis for subsequent TDOA time difference positioning. Based on this, the swarm conducts wide-area distributed cruise search in the target area according to a preset coverage path (such as a bow-shaped or randomly scattered pattern). Each UAV continuously monitors the spectrum environment, focusing on finding preset characteristics of specific target signal sources (such as sudden changes in RSSI received signal strength at a specific frequency), thereby efficiently identifying suspected radiation sources.

[0041] As a preferred embodiment, real-time calibration and correction of the frequency and phase of the local crystal oscillator specifically includes: S31. Receive BeiDou pulses and time information sent by BeiDou satellites, and use the BeiDou pulses as an absolute time reference.

[0042] S32. Divide the high-frequency clock signal output by the local crystal oscillator to generate a local pulse.

[0043] S33. The BeiDou pulse and the local pulse are compared by a phase detector to determine the time difference between their rising edges, and the phase difference signal is output.

[0044] S34. The phase difference signal is processed by a loop filter to generate a control voltage.

[0045] S35. Apply the control voltage to the control terminal of the voltage-controlled crystal oscillator to adjust the frequency and phase of the local crystal oscillator in real time.

[0046] In this embodiment, the airborne BeiDou timing module receives 1PPS (pulse per second) and TOD (time information) signals transmitted by BeiDou satellites. 1PPS is a pulse signal with an extremely precise rising edge (error on the nanosecond level), representing the instant at the start of each second. Internally, the module also generates a local second pulse from the clock signal generated by a local high-stability crystal oscillator (such as a temperature-controlled crystal oscillator OCXO) through a frequency divider circuit. By comparing the phase difference between these two second pulses, the control voltage of the local crystal oscillator is continuously adjusted to gradually align the local second pulse with the BeiDou second pulse, thereby achieving real-time frequency and phase calibration.

[0047] Specifically, after the BeiDou receiver locks onto a satellite, it outputs two key signals: 1PPS and TOD. 1PPS is a pulse per second, with the absolute timing of its pulse edge being highly accurate (corresponding to whole seconds in UTC time). TOD is serial time data, providing the absolute time (year, month, day, hour, minute, second) corresponding to the current 1PPS. The onboard high-stability crystal oscillator (OCXO) outputs a high-frequency sine wave or square wave (e.g., 10MHz or 100MHz). The output signal is fed into a digital frequency divider, with the division factor equal to the crystal oscillator frequency (e.g., 10MHz divided to 1Hz). The output of the divider is the local second pulse. Ideally, if the crystal oscillator frequency is absolutely accurate, the local second pulse should be perfectly aligned with the BeiDou 1PPS. However, in reality, due to factors such as initial frequency error, temperature drift, and aging of the crystal oscillator, the local second pulse will gradually lead or lag behind.

[0048] Two second pulses (BeiDou 1PPS and a local second pulse) are fed into a phase detector (which can be a time-to-digital converter (TDC) or a simple edge-triggered counter). The phase detector detects the time difference (i.e., phase difference) between the rising edges of the two pulses and outputs a voltage or digital value proportional to the time difference. For example, if the local second pulse lags behind the BeiDou 1PPS by 10 nanoseconds, a positive error signal is output; if it leads, a negative error signal is output. The phase difference signal output by the phase detector is fed into a loop filter (usually a low-pass filter, sometimes combined with a PID control algorithm). The loop filter smooths short-term noise, extracts the long-term average deviation, and generates a stable control voltage. This control voltage is converted into an analog voltage by a digital-to-analog converter (DAC) and applied to the voltage-controlled oscillator (VCXO) or OCXO. The oscillation frequency of the OCXO changes linearly with the control voltage (voltage control sensitivity, such as ±1ppm / V).

[0049] If the local second pulse lags (phase difference is positive), the control voltage increases, slightly increasing the OCXO frequency to accelerate the local clock and reduce lag; if it leads, the control voltage decreases, the frequency decreases, and the local clock slows down. For high-stability crystal oscillators, a simple phase-locked loop may not be able to fully compensate for long-term aging drift. In this case, a discipline algorithm (such as Kalman filtering, least squares fitting, or adaptive PID) is added. The algorithm records historical phase difference data over a period of time (e.g., several hours or days), statistically fits the frequency shift trend and aging rate of the crystal oscillator, and then predicts and outputs a better compensation control quantity. This allows for short-term high-precision timekeeping by relying on the memorized discipline parameters even when the satellite signal is briefly lost (e.g., entering a tunnel).

[0050] The aforementioned comparison-filtering-control-feedback process repeats every second (or at a higher frequency, such as multiple times per second). After several adjustment cycles, the phase difference between the local second pulse and the BeiDou 1PPS is locked within a very small range (e.g., ±5 nanoseconds). At this point, the frequency and phase of the OCXO are "tamed" to synchronize with the BeiDou satellite atomic clock. Finally, through real-time tracking by the phase-locked loop, the output frequency of the local OCXO is locked to the BeiDou reference frequency, eliminating clock deviations and drift between the various UAV nodes. All UAVs follow the same BeiDou space-based reference, achieving clock synchronization accuracy at the nanosecond level, providing a unified timestamp basis for subsequent TDOA (Time Difference of Arrival) positioning.

[0051] As a preferred embodiment, determining the target region of the target signal source specifically includes: S41. When any one or more drones detect that the strength of a target signal exceeds a preset threshold, it is determined that a suspected target signal has been detected, and a trigger signal is broadcast to nearby drones through a self-organizing communication network.

[0052] S42. Upon receiving the trigger signal, nearby drones move closer to the triggering drone, forming a local high-density cluster around the suspected target signal.

[0053] S43. Each UAV in the cluster measures the signal-to-noise ratio of the target signal it receives and aggregates the signal-to-noise ratio information to the server.

[0054] S44. Based on the spatial distribution of the signal-to-noise ratio, the server determines the location where the signal-to-noise ratio intensity is lower than a preset threshold as the signal coverage edge.

[0055] S45. The UAV located at the edge of the signal coverage area hovers or scans, dynamically measures the signal boundary, gradually delineates the coverage outline of the signal source, and identifies the area inside the outline as the target area where the target signal source may exist.

[0056] In this embodiment, an intelligent event-triggered mechanism is used to shift from blind wide-area search to targeted collaborative detection. When any one or more drones (denoted as node X) detects that the strength of a target signal exceeds a preset threshold, it is determined that a suspected target has been detected. Node X immediately sends a trigger signal through its ad hoc network to all neighboring nodes that can directly receive the broadcast signal. The broadcast content includes information such as the center frequency and bandwidth of the target signal, so that neighboring nodes can identify whether they have detected the same target and activate the collaborative detection mode. Upon receiving the trigger signal, the neighboring drones then move towards node X, forming a local high-density cluster around the suspected target.

[0057] Within the cluster, each UAV communicates with the others, aggregating their measured signal-to-noise ratio (SNR) data to a ground server. The server then determines the edge of the signal coverage based on the spatial distribution of the SNR. Locations with weak SNR are identified as the signal coverage edge, while locations with strong SNR are identified as the core area. The system further instructs UAVs located at the signal edge to hover or scan, dynamically outlining the coverage contour of the signal source, thereby continuously narrowing down the area where the target might exist. The cooperative boundary continues to expand outward until the target can no longer be detected; the location where the target can no longer be detected is determined as the final boundary of the signal coverage.

[0058] As a preferred embodiment, calculating the virtual center point of the target region specifically includes: S51. Obtain the position coordinates of all UAVs participating in the boundary exploration and the received signal-to-noise ratio of the same target signal measured by each UAV.

[0059] S52. Use the received signal-to-noise ratio (SNR) value of each drone as the weight of each drone in the calculation, and add up the SNR weights of all drones to obtain the total weight.

[0060] S53. Multiply the longitude, latitude, and altitude values ​​of each UAV by their corresponding signal-to-noise ratio weights to obtain the weighted longitude, weighted latitude, and weighted altitude of each UAV. Sum the weighted longitude, weighted latitude, and weighted altitude of all UAVs to obtain the total weighted longitude, total weighted latitude, and total weighted altitude.

[0061] S54. Divide the total weighted longitude by the total weight to obtain the longitude value of the virtual center point; divide the total weighted latitude by the total weight to obtain the latitude value of the virtual center point; divide the total weighted altitude by the total weight to obtain the altitude value of the virtual center point.

[0062] S55. By integrating the longitude, latitude, and altitude values, the three-dimensional coordinates of the virtual center point are obtained.

[0063] In this embodiment, the virtual center point of the target area is obtained by calculating the weighted centroid using the coordinates of all boundary exploration drones and the signal-to-noise ratio (SNR) of the target signal received by each drone. Specifically, using a weighted centroid algorithm, the system obtains the position coordinates (e.g., latitude and longitude or spatial rectangular coordinates) of all drone nodes participating in boundary exploration, as well as the received SNR of the same target signal measured by each drone. The SNR value of each drone is directly used as the "weight" of that node in the calculation. The higher the SNR, the better the signal quality received by the node, and the greater its influence on the center point position. The system performs weighted summation on each dimension of the coordinates. Specifically, the longitude value (or X-coordinate) of each drone is multiplied by its SNR weight to obtain the weighted longitude of that node; the weighted longitudes of all nodes are summed to obtain the total weighted longitude.

[0064] Similarly, the latitude and altitude (or Y and Z coordinates) are weighted and summed in the same way; simultaneously, the signal-to-noise ratio (SNR) weights of all UAVs are summed to obtain the total weight. The total weighted longitude is divided by the total weight to obtain the longitude value of the virtual center point; the total weighted latitude is divided by the total weight to obtain the latitude value of the virtual center point; if altitude is involved, the altitude value is calculated similarly. That is, the longitude, latitude, and altitude values ​​constitute the complete coordinates of the virtual center point. The calculated virtual center point will naturally be biased towards areas with high SNR. For example, if the SNR of a certain node is twice that of other nodes, then that node acts as two ordinary nodes in the same position in the coordinate calculation, thus pulling the center point towards that node. Finally, the obtained virtual center point coordinates are output as the reference coordinates for subsequent selection of reference nodes.

[0065] As a preferred embodiment, the determination of spatial geometric relationships specifically includes: S61. Establish a local three-dimensional coordinate system with the virtual center point of the target area as the origin.

[0066] S62. Obtain the position coordinates and attitude information of each UAV participating in the exploration in the global coordinate system.

[0067] S63. Using the direction cosine matrix, the position coordinates and attitude information are transformed into a local three-dimensional coordinate system to obtain the three-dimensional direction vector of each UAV relative to the virtual center point.

[0068] S64. Calculate the axial proximity between the three-dimensional direction vector of each UAV and the three orthogonal coordinate axes in the local three-dimensional coordinate system.

[0069] S65. Based on axial proximity, the UAV closest to the three orthogonal axes is selected as the positioning UAV, and the spatial geometric relationship between the positioning UAV and the target signal source is determined.

[0070] In this embodiment, after determining the virtual center point of the target area, the system uses this center point as the origin and, using the attitude and position information provided by the airborne BeiDou module, transforms the positions of each UAV in the global ECEF coordinate system to a local coordinate system with the center point as the origin using a direction cosine matrix. It then calculates the three-dimensional direction vector of each node relative to the center point and identifies the UAV node closest to the three orthogonal axes (e.g., X-axis, Y-axis, Z-axis, or 0°, 90°, and 180° directions on the horizontal plane). The system does not include all participating UAVs in the final calculation. Instead, based on the principle of minimizing the geometrical precision factor (GDOP), it selects 3 to 4 UAVs that can form the optimal spatial geometry as "positioning reference UAVs," thus ensuring that the intersection angle of the positioning hyperbola equations is close to the optimal state, laying a good geometric foundation for subsequent high-precision solutions.

[0071] Specifically, by transforming the positions of each UAV to a local coordinate system with the center point as the origin, calculating its direction vector, and selecting the node closest to the three orthogonal axes, the positioning reference UAV with the best spatial geometric relationship is determined. First, a local three-dimensional coordinate system (e.g., the X, Y, and Z axes are mutually orthogonal) is established with the virtual center point calculated by the weighted centroid algorithm as the origin. Each participating UAV obtains its own position coordinates and attitude information in the global geocentric-fixed coordinate system (ECEF) through its onboard BeiDou module. Then, using the direction cosine matrix, the global coordinates of each UAV are transformed to a local coordinate system with the center point as the origin. After the transformation, each node obtains a three-dimensional direction vector relative to the center point (i.e., the vector pointing from the center point to the UAV).

[0072] For each node, the angle or cosine value between its direction vector and the three orthogonal coordinate axes (e.g., positive X-axis, positive Y-axis, and positive Z-axis) is calculated, and the UAV node closest to each axis (with the smallest angle) is identified. Finally, the system selects 3-4 UAVs located in the three orthogonal directions (or approximately orthogonal) from all candidate nodes as "positioning reference UAVs". These nodes can form the minimum geometrical precision factor (GDOP), i.e., the optimal spatial geometry. Determining the spatial geometry of this set of UAV nodes, which form an approximately orthogonal axial distribution relative to the center point, provides the best observation geometry for subsequent time difference positioning or angle measurement.

[0073] As a preferred embodiment, determining the three-dimensional coordinates of the target signal source specifically includes: S71. Simultaneously intercept signals emitted by the target signal source by positioning the drone, and record the arrival timestamp of each intercepted signal.

[0074] S72. Calculate the initial position of the target signal source based on the arrival timestamp.

[0075] S73. Iteratively correct the initial position to obtain the three-dimensional coordinates of the target signal source.

[0076] In this embodiment, the system uses a selected positioning reference UAV for final calculation. First, the selected reference UAV simultaneously intercepts the target signal and records the precise arrival timestamp of each signal using a unified high-precision time reference; each node sends the measurement data with high-precision timestamps to the main processing node (which can be the lead UAV or a ground station) through an ad hoc network.

[0077] In the algorithm solution stage, the system first employs the Chan algorithm as the core algebraic solution method for coarse positioning. By introducing auxiliary variables, the originally nonlinear TDOA hyperbolic equation system is transformed into a pseudo-linear equation system, which is then directly solved using the two-step weighted least squares (WLS) method. This allows for the rapid acquisition of the initial estimated position of the target signal source without iteration, meeting the stringent real-time requirements of UAV swarms. Furthermore, its positioning accuracy approaches the Cramer-Rao lower bound under moderate noise levels, exhibiting better robustness to geometric accuracy factors than traditional methods and avoiding the local convergence problem that may exist in iterative algorithms. In addition, the Chan algorithm provides a stable and accurate initial estimated position for subsequent Taylor series fine iterations, thus achieving both real-time performance and reliability of the overall positioning scheme while ensuring accuracy.

[0078] Subsequently, using this initial estimated position as the initial value, the Taylor series expansion method is employed for precise positioning iterative correction. In each iteration, the current estimated position is substituted into a system of nonlinear equations to recalculate the theoretical time difference (TDOA), and the error is compared with the measured TDOA. If the error decreases, the correction value is accepted; otherwise, the iteration continues, gradually eliminating measurement errors and approximating the true position. Finally, the system outputs the precise three-dimensional coordinates of the target signal source. In TDOA positioning, the target position and the measured distance difference (converted from the signal arrival time difference multiplied by the speed of light) exhibit a strongly nonlinear relationship. That is, when the target moves in space, the distance difference received by each UAV does not change linearly, thus it cannot be directly solved using a system of linear equations. The core idea of ​​the Taylor series method is to approximate the nonlinear relationship in the vicinity of the current estimated position using a first-order approximation, i.e., using a "tangent line" or "tangent plane" at the estimated point to locally replace the original curve or surface.

[0079] Specifically, the distance difference function between the actual target and each UAV can be decomposed into the theoretical distance difference at the current estimated position, plus a first-order term consisting of the position correction multiplied by the rate of change of the distance difference at that point (i.e., the directional derivative or geometric gradient). Moving the theoretical distance difference to one side of the equation yields an approximate linear relationship between the measurement residual and the position correction. Thus, a linear equation for the three directional corrections can be written for each reference UAV, and all equations combined form a linear overdetermined system of equations. The corrections can be directly solved using the least squares method, avoiding the difficulty of directly solving complex nonlinear equations. Through multiple iterations, each time the same linear approximation is performed at the updated position, the true target position can be gradually approximated. That is, each iteration only requires solving the linear equation system once, resulting in high computational efficiency, fast iteration convergence, and controllable error of the first-order approximation.

[0080] As a preferred embodiment, calculating the initial position of the target signal source specifically includes: S81. Obtain the three-dimensional coordinates of at least three reference UAVs and determine the reference UAV.

[0081] S82. Obtain the arrival time difference of the target signal source to each reference UAV relative to the reference UAV, and convert the time difference into a distance difference.

[0082] S83. Introduce an auxiliary variable, which is the distance from the target signal source to the reference UAV.

[0083] S84. Using the distance difference and auxiliary variables, the nonlinear TDOA hyperbolic equation system is transformed into a pseudo-linear equation system with respect to the three-dimensional coordinates of the target signal source and auxiliary variables.

[0084] S85. The pseudo-linear equation system is solved using the first weighted least squares method to obtain the initial solutions for the target coordinates and auxiliary variables.

[0085] S86. Using the geometric constraint relationship between the three-dimensional coordinates of the target signal source and the auxiliary variables, construct the second weighted least squares equation, correct the initial solution, and obtain the initial position of the target signal source.

[0086] In this embodiment, the position coordinates of at least three selected reference UAVs and the arrival time difference (converted to distance difference) measured with one of them as a reference node are first used as input. The core of the algorithm lies in introducing an auxiliary variable—the distance from the target to the reference UAV—thereby transforming the nonlinear hyperbolic equation system into a pseudo-linear equation system about the target coordinates and this auxiliary variable.

[0087] In practice, a linear equation is first constructed for each non-reference node, and all equations are combined to form a matrix. Then, a first weighted least squares method is used to solve the problem, with the weighting matrix set based on the covariance matrix of the TDOA measurements, yielding an initial solution for the target coordinates and auxiliary variables. Since this initial solution does not satisfy the geometric constraints between the auxiliary variables and the target coordinates (i.e., the distance from the target to the reference UAV should be equal to the Euclidean distance between the two points), the algorithm performs a second weighted least squares: using this geometric constraint, a new system of linear equations is constructed, the solution from the first step is projected into the constraint space, and the solution is solved again to minimize the constraint residuals. The target coordinates output after two weighted least squares steps are the coarse positioning result. This method requires no iteration, is computationally fast, and can be directly used as the initial value for subsequent Taylor series iterations.

[0088] As a preferred embodiment, the initial position is iteratively corrected, specifically including: S91. Based on the initial position, construct a system of linear equations about the position correction using Taylor series, and solve them to obtain the correction amount for the current position.

[0089] S92. Add the correction amount to the initial position to obtain the candidate new position, and calculate the error index of the candidate new position.

[0090] S93. If the error index of the candidate new position is less than the error index of the initial position, then the candidate new position is accepted as the estimated position for the next iteration; otherwise, the candidate position is recalculated.

[0091] S94. When the correction amount reaches the preset iteration condition, stop the iteration and output the three-dimensional coordinates of the target signal source.

[0092] In this embodiment, the Taylor series iterative correction method uses the initial estimated position output by the Chan algorithm as the initial value for iteration, and inputs the three-dimensional coordinates of at least four preferred reference UAVs and the measured distance difference between the target signal and each non-reference node relative to the reference node (obtained by multiplying TDOA by the speed of light). Specifically, the target's true position is considered as the current estimated position plus a small correction amount. Each iteration first calculates the theoretical distance from the target to the reference UAV and to each of the other reference UAVs at the current estimated position, thus obtaining the theoretical distance differences. Then, the residual between the measured distance difference and the theoretical distance difference is calculated. Next, the Taylor first-order expansion is used to express the residual as a linear function of the position correction amount, where the coefficients are determined by the difference in direction cosines between the current estimated position and each UAV, thereby establishing a linear overdetermined system of equations about the correction amount.

[0093] The weighted least squares method (with the weight matrix determined by the TDOA measurement covariance matrix) is used to solve the system of equations to obtain the correction for the current step. Candidate new positions are then generated, and the error index (e.g., the sum of squared residuals) at each candidate position is calculated. The old and new errors are compared: if the new error is smaller, the new position is accepted as the estimated position for the next iteration; otherwise, the correction is multiplied by a scaling factor less than 1, and the iteration is repeated until the error decreases or the maximum number of iterations is reached. After each iteration, the convergence condition is checked: the correction magnitude is less than a preset threshold (e.g., 1 meter), the error rate of change is less than a threshold (e.g., 1%), or the maximum number of iterations is reached (e.g., 10 iterations). If any of these conditions are met, the iteration stops, and the current estimated position is output as the refined target 3D coordinates.

[0094] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A radiation source localization system based on swarm drone cooperation and ad hoc network communication, characterized in that, The system includes: The signal frequency acquisition module is used to build a self-organizing communication network for the target area through a swarm of drones and continuously monitor and acquire preset characteristics of the target signal source; The signal edge determination module is used to determine the target area of ​​the target signal source by outlining the contour of the signal coverage area when any drone detects the preset characteristics of the target signal source. The target node determination module is used to calculate the virtual center point of the target area based on the positions of all UAVs participating in the boundary exploration, and to determine the positioning UAVs located in three orthogonal directions with the virtual center point as the origin, and to determine the spatial geometric relationship between the positioning UAVs and the target signal source. The signal coordinate determination module is used to determine the three-dimensional coordinates of the target signal source based on spatial geometric relationships.

2. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, Constructing a self-organizing communication network for a target environment using drone swarms specifically includes: S11. Each drone is equipped with a communication module, which loads preset network parameters and initializes the network protocol stack after power-on. S12. Each UAV periodically broadcasts a beacon frame containing its own identifier and status on a preset channel, while simultaneously listening to the beacon frames of other UAVs. S13. Upon receiving the beacon frame, the drone records the identifier and signal strength of the sending drone and generates a neighbor table; S14. Each UAV exchanges routing information based on its neighbor table and through a self-organizing routing protocol, dynamically establishing a routing table and forming a self-organizing communication network. S15. During network operation, each UAV continuously monitors the link status of the self-organizing communication network. When a link interruption or node change is detected, route repair and topology reconstruction are automatically triggered.

3. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, Each drone is equipped with a BeiDou positioning and timing unit to collect the signal frequency of the target signal source, specifically including: S21. Each UAV receives the time signal from the Beidou satellite and calibrates and corrects the frequency and phase of the local crystal oscillator in real time to obtain the real-time position coordinates of each UAV. S22. The drone swarm conducts a distributed wide-area search in the target area according to the preset coverage path. Each drone continuously monitors the spectrum environment of the target area and detects the preset characteristics of the target signal source in real time.

4. The radiation source localization system based on swarm UAV collaboration and ad hoc network communication as described in claim 3, characterized in that, Real-time calibration and correction of the local crystal oscillator's frequency and phase, specifically including: S31. Receive BeiDou pulses and time information sent by BeiDou satellites, and use BeiDou pulses as absolute time references; S32. Divide the high-frequency clock signal output by the local crystal oscillator to generate a local pulse; S33. Compare the time difference between the rising edges of the BeiDou pulse and the local pulse using a phase detector, and output the phase difference signal. S34. After processing the phase difference signal through a loop filter, a control voltage is generated; S35. Apply the control voltage to the control terminal of the voltage-controlled crystal oscillator to adjust the frequency and phase of the local crystal oscillator in real time.

5. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, Determining the target region of the target signal source specifically includes: S41. When any one or more drones detect that the strength of a target signal exceeds a preset threshold, it is determined that a suspected target signal has been detected, and a trigger signal is broadcast to nearby drones through a self-organizing communication network. S42. Nearby drones that receive the trigger signal move toward the triggering drone, forming a local high-density cluster around the suspected target signal. S43. Each UAV in the cluster measures the signal-to-noise ratio of the target signal it receives and aggregates the signal-to-noise ratio information to the server. S44. The server determines the location where the signal-to-noise ratio is lower than a preset threshold as the signal coverage edge based on the spatial distribution of the signal-to-noise ratio. S45. The UAV located at the edge of the signal coverage area hovers or scans, dynamically measures the signal boundary, gradually delineates the coverage outline of the signal source, and identifies the area inside the outline as the target area where the target signal source may exist.

6. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, Calculating the virtual center point of the target region specifically includes: S51. Obtain the position coordinates of all drones participating in the boundary exploration and the received signal-to-noise ratio of the same target signal measured by each drone; S52. Use the received signal-to-noise ratio value of each drone as the weight of each drone in the calculation, and add up the signal-to-noise ratio weights of all drones to obtain the total weight. S53. Multiply the longitude, latitude, and altitude values ​​of each UAV by the corresponding signal-to-noise ratio weights to obtain the weighted longitude, weighted latitude, and weighted altitude of each UAV. Sum the weighted longitude, weighted latitude, and weighted altitude of all UAVs to obtain the total weighted longitude, total weighted latitude, and total weighted altitude. S54. Divide the total weighted longitude by the total weight to obtain the longitude value of the virtual center point; divide the total weighted latitude by the total weight to obtain the latitude value of the virtual center point; divide the total weighted altitude by the total weight to obtain the altitude value of the virtual center point. S55. By integrating the longitude, latitude, and altitude values, the three-dimensional coordinates of the virtual center point are obtained.

7. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, The determination of spatial geometric relationships specifically includes: S61. Establish a local three-dimensional coordinate system with the virtual center point of the target area as the origin; S62. Obtain the position coordinates and attitude information of each UAV participating in the exploration in the global coordinate system; S63. Using the direction cosine matrix, the position coordinates and attitude information are transformed into a local three-dimensional coordinate system to obtain the three-dimensional direction vector of each UAV relative to the virtual center point. S64. Calculate the axial proximity between the three-dimensional direction vector of each UAV and the three orthogonal coordinate axes in the local three-dimensional coordinate system; S65. Based on axial proximity, the UAV closest to the three orthogonal axes is selected as the positioning UAV, and the spatial geometric relationship between the positioning UAV and the target signal source is determined.

8. The radiation source localization system based on swarm drone collaboration and ad hoc network communication as described in claim 1, characterized in that, Determining the three-dimensional coordinates of the target signal source specifically includes: S71. Simultaneously intercept signals emitted by the target signal source by positioning the drone, and record the arrival timestamp of each intercepted signal; S72. Calculate the initial position of the target signal source based on the arrival timestamp; S73. Iteratively correct the initial position to obtain the three-dimensional coordinates of the target signal source.

9. The radiation source localization system based on swarm UAV collaboration and ad hoc network communication as described in claim 8, characterized in that, Calculating the initial position of the target signal source specifically includes: S81. Obtain the three-dimensional coordinates of at least three reference UAVs and determine the reference UAV; S82. Obtain the arrival time difference of the target signal source to each reference UAV relative to the reference UAV, and convert the time difference into a distance difference; S83. Introduce an auxiliary variable, which is the distance from the target signal source to the reference UAV; S84. Using distance difference and auxiliary variables, the nonlinear TDOA hyperbolic equation system is transformed into a pseudo-linear equation system with respect to the three-dimensional coordinates of the target signal source and auxiliary variables. S85. The pseudo-linear equation system is solved using the first weighted least squares method to obtain the initial solution of the target coordinates and auxiliary variables; S86. Using the geometric constraint relationship between the three-dimensional coordinates of the target signal source and the auxiliary variables, construct the second weighted least squares equation, correct the initial solution, and obtain the initial position of the target signal source.

10. The radiation source localization system based on swarm UAV collaboration and ad hoc network communication as described in claim 8, characterized in that, The initial position is iteratively corrected, specifically including: S91. Based on the initial position, construct a system of linear equations about the position correction using Taylor series, and solve them to obtain the correction amount for the current position. S92. Add the correction amount to the initial position to obtain the candidate new position, and calculate the error index of the candidate new position; S93. If the error index of the candidate new position is less than the error index of the initial position, then accept the candidate new position as the estimated position for the next iteration; otherwise, recalculate the candidate position. S94. When the correction amount reaches the preset iteration condition, stop the iteration and output the three-dimensional coordinates of the target signal source.

Images